Minimization of Artifacts in Sulfuric Acid Mist Measurement Using

Jun 24, 2008 - University of Florida, Gainesville, Florida 32611-6450, and. Florida Institute of Phosphate Research,. Bartow, Florida 33830. Received ...
0 downloads 5 Views 521KB Size
Environ. Sci. Technol. 2008, 42, 5694–5699

Minimization of Artifacts in Sulfuric Acid Mist Measurement Using NIOSH Method 7903 Y U - M E I H S U , † C H A N G - Y U W U , * ,† DALE A. LUNDGREN,† AND BRIAN K. BIRKY‡ Department of Environmental Engineering Sciences, University of Florida, Gainesville, Florida 32611-6450, and Florida Institute of Phosphate Research, Bartow, Florida 33830

Received February 19, 2008. Revised manuscript received April 8, 2008. Accepted May 9, 2008.

NIOSH Method 7903 employs a silica gel tube for sulfuric acid mist measurement in workplaces. However, SO2 gas present in the sample volume can be transformed into sulfate in the sampling process causing an artifact that is reported as sulfuric acid. A sampling train incorporating a honeycomb denuder system was applied for field sampling at seven phosphate fertilizer plants to evaluate its use for reducing the artifact sulfate concentration while preserving the actual sulfuric acid mist concentration. The denuder system was designed to remove SO2 gas before the air entered the silica gel tube and to monitor SO2 concentration at the same time. A deactivation model was also applied to correct for the presence of the artifact. The denuder system had 95.7 ( 6.8% collection efficiency for SO2 gas, and the impact of sulfate aerosol on SO2 collection was negligible. SO2 concentrations at the seven plants ranged from 34 ppb to 5.6 ppm. The honeycomb denuder system and the deactivation model were shown to reduce the artifact sulfate concentration by 70% and 39%, respectively. However, they were still higher than the sulfate aerosol concentration measured by a cascade impactor. One possible reason is the residual sulfate in the glass fiber filter and the silica gel.

Introduction NIOSH (National Institute for Occupational Safety and Health) Method 7903 (1) is an approved method set by the Occupational Safety and Health Administration (OSHA) for measuring the total concentration of acidic aerosols and gases, including HF, HCl, HBr, HNO3, H2SO4, and H3PO4. It is the method commonly used by health and safety staff in the phosphate industry (2) as well as other occupational environments such as the semiconductor industry, lead battery factories, aluminum smelting, machining, electroplating processes, and even disaster response (3–6). The sampler used for NIOSH Method 7903, a silica gel tube, consists of one section of glass fiber filter plug followed by two sections of silica gel. The glass fiber filter plug is designed to filter out the majority of aerosols while the silica gel sections are used mainly to adsorb acidic gases. The NIOSH recommended sampling flow rate range is 0.2-0.5 Lpm, except * Corresponding author tel: (352) 392-0845; fax: (352) 392-3076; e-mail: [email protected]. † University of Florida. ‡ Florida Institute of Phosphate Research. 5694

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 15, 2008

that less than 0.3 Lpm should be used for HF. The collected samples are desorbed in eluent and the aliquots are analyzed by ion chromatography (IC). In evaluating the method, NIOSH researchers reported nearly 100% collection efficiency for acidic gases (HCl, HF, and HNO3) (7, 8). For aerosols (H2SO4 and H3PO4), around 90% efficiency (94.8 ( 4.8% for H3PO4 and 86 ( 4.6% for H2SO4) was reported when the samples collected on the glass fiber filter section and the front silica gel section were combined. Ortiz and Fairchild (9) reported approximately 70% of the aerosol mass was collected by the glass fiber filter plug although the efficiency varied depending on the size distribution of the sampling aerosol (10). Sulfur dioxide (SO2) is a species which can be adsorbed by the glass fiber filter (11–14) and the silica gel (15–17). Once adsorbed, it is subsequently transformed into sulfate in the analytical process that causes an overestimate of sulfuric acid mist concentration (18). Hsu et al. (18) carried out experiments in a laboratory system that verified and quantified the artifacts resulting from the adsorption and conversion of interfering SO2 into sulfate. A deactivation model (16) was modified for estimating the artifact sulfate concentration based on the various SO2 concentrations, flow rates, and sampling times. Denuder systems have been widely used for air sampling (19–24). Various types of denuders have been developed and are commercially available. The denuder’s wall is coated with pertinent adsorbents depending on the gaseous species of interest. When air passes through the denuder, gas molecules with large diffusivity can diffuse to the wall and get adsorbed. Basic adsorbent (e.g., sodium carbonate (Na2CO3)/glycerol) can be used for acidic gases, such as HCl, HNO3, SO2, and HNO2. Aerosols can be collected in the filter that follows. As a consequence, this design can reduce the interference of gaseous species on aerosols. The removal efficiency of SO2 using an annular denuder system, depending on the operating flow rate, has been reported to be >99% when operated at a low flow rate (25). The denuder system (26) has also been applied for ambient air sampling and can provide high SO2 collection efficiency. Accurate determination of sulfuric acid mist concentration in fertilizer manufacturing facilities is seminal to the evaluation of its occupational exposure. Minimizing the artifact due to SO2 gas in the sampling process is essential in reaching this goal. The objectives of this study are 2-fold: (1) to explore the use of a denuder for removing SO2 gas from the sampling volume and, (2) to assess the applicability of the deactivation model for correcting the artifacts for the phosphate fertilizer production environment.

Materials and Methods Two methods, a sampling system cooperating with a honeycomb denuder system (HDS) and a deactivation model, were applied to minimize the artifact sulfate. Field sampling was carried out to examine the artifact removal efficiency when the HDS was applied to remove the interfering SO2 before entering the standard sampling train. Experiments were conducted to characterize SO2 adsorption at high SO2 concentrations encountered in the field sampling conditions. The deactivation model was used to estimate the artifact sulfate concentration based on known SO2 concentrations. Field Sampling. Field sampling was carried out on top of the sulfuric acid pump tank area at seven phosphate fertilizer plants in Florida (27). Four samples were acquired at each of six sites and five samples were collected at the one remaining plant. Sampling time was 8 h for each sample. 10.1021/es800502u CCC: $40.75

 2008 American Chemical Society

Published on Web 06/24/2008

TABLE 1. Statistics of SO2 Concentrations (ppm)

FIGURE 1. Three sampling trains: (1) NIOSH Method 7903, (2) modified NIOSH Method 7903, (3) cascade impactor sampling system. Silica gel tubes, a cascade impactor, and a honeycomb denuder system were applied for the field sampling. Three sets of sampling trains, shown in Figure 1, were employed. (1) A silica gel tube: This method was used for total sulfuric acid mist concentration following NIOSH Method 7903 (N ) 29). (2) The HDS followed by two silica gel tubes in parallel: The HDS (coating solution: 1% Na2CO3/glycerol) was used to remove SO2 gas before the air entered the silica gel tubes. The measurement also provided SO2 concentration. Two silica gel tubes were applied for replication (N ) 29 for the HDS sample; N ) 58 for the silica gel tube sample). (3) A cascade impactor (Mark III, U. Washington) with Zefluor membrane filters (P5PJ001, Pall Corp.): This sampling system was used for sulfuric acid mist concentration with sizeresolved information (N ) 29). Zefluor membrane filters can provide high aerosol collection efficiency with low interaction with acidic gases (12, 14). All three sampling trains were set side by side for parallel sampling. All silica gel tube samples were analyzed following the sample preparation procedures outlined in NIOSH Method 7903 (1), including adding 9 mM Na2CO3 of 10 mL and a water bath at 100 °C for 10 min. The HDS sample was extracted by 10 mL of deionized (DI) water (Nanopure Diamond, Barnstead), and 1 mL of hydrogen peroxide (0.6%) was then added into 1 mL of sample solution to oxidize sulfite to sulfate for the analysis. The cascade impactor sample was extracted by 10 mL of DI water with a 1-h ultrasonic bath. Sulfate concentrations of all samples were analyzed via IC (model ICS 1500, Dionex). Deactivation Model. A deactivation model (18), eq 1, has been developed to estimate the artifact sulfate concentration. In developing this model, experiments were conducted at SO2 concentrations ranging from 0.2 to 1.6 ppm. Hence, the model can be applied for SO2 concentrations in this range.

{

[SO24 ]t ) θ[SO2]o

[

]}

ksSo exp(-kdt) Q for 0.2 ppm < [SO2]o < 1.6 ppm (1)

1 - exp -

3 where [SO24 ]t is artifact sulfate concentration (µg/m ); θ ) 1571, conversion factor; [SO2]o ) feed SO2 concentration

site

A

B

C

D

E

F-1

F-2

G

mean min max sample no.

1.83 0.90 2.57 4

1.74 1.22 2.66 4

0.46 0.36 0.56 4

0.05 0.03 0.05 4

0.27 0.19 0.31 4

0.40 0.12 0.68 2

5.00 4.34 5.64 3

0.21 0.13 0.31 4

(ppm); ksSo ) observed adsorption rate (mLpm/min), 16 for the glass fiber filter and 8.9 for the silica gel at the flow rate, Q, of 500 mLpm; kd ) first order deactivation rate constant (min-1), 0.0026 for the glass fiber filter and 0.0014 for the silica gel at the flow rate of 500 mLpm; and t ) time (min). SO2 Adsorption. High SO2 concentrations, 1.6 ppm to 5.6 ppm, were observed during the field sampling described in the next section. However, the parameter values in the deactivation model had not been validated for the artifact estimate at high SO2 concentrations. Therefore, experiments were conducted to quantify the SO2 adsorption by the silica gel tube under high SO2 concentrations ranging from 1.6 to 5.6 ppm at 0.5 Lpm flow rate with 8 h of sampling time (condition employed for the field sampling). The experimental setup has been described in Hsu et al. (18). SO2 gas from a cylinder (10 ppm, relative uncertainty ) (5%) was mixed with zero air (model 111, Thermo Electron Instrument) to obtain the desired concentration. SO2 gas was then passed through the silica gel tube for the adsorption. An impinger with 100 mL of 9 mM Na2CO3 solution was employed downstream to collect the residual SO2 gas. Sulfate concentration was determined following the analytical procedures described in the field sampling section. To evaluate residual sulfate of both glass fiber filter and silica gel, 18 silica gel tubes as received were analyzed for their sulfate concentration following the sample preparation procedures of NIOSH Method 7903.

Results and Discussion Field Sampling. SO2 Collection Efficiency and SO2 Concentration. The HDS consists of two honeycomb denuders in series. The collection efficiency, η, was determined according to eq 2: η(%) )

[SO2(HD1)] - [SO2(HD2)] × 100% [SO2(HD1)]

(2)

where [SO2(HD1)] and [SO2(HD2)] are SO2 concentrations collected by the first and second honeycomb denuders, respectively. The mean SO2 collection efficiency ( standard deviation of the HDS was found to be 95.7 ( 6.8% (N ) 29), demonstrating that the deployed HDS could effectively remove SO2 gas from the sample gas. Regarding the SO2 concentration at the phosphate fertilizer plants, it ranged from 34 ppb to 5.6 ppm. Although it varied significantly from plant to plant, for each plant the SO2 concentration level varied within a limited range during the sampling period. Table 1 displays the statistics of SO2 concentrations at each plant. At plant F, sampling was conducted at 2 locations where SO2 concentrations differed greatly. Hence, SO2 concentrations at plant F are shown separately for each location. Ratio of S-SO42-/S-SO2. The deployment of the HDS was to collect SO2 gas; however, sulfate aerosols might also be collected by the HDS which could cause an overestimate of SO2 concentration. To determine if the presence of sulfate significantly affects the SO2 concentration measurement, the S-SO42-/S-SO2 ratios (elemental sulfur from SO42- collected by the silica gel tubes, SGHA and SGHB in the second sampling train, to elemental sulfur from SO2 collected by the HDS) for all samples were calculated and are shown in Figure VOL. 42, NO. 15, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5695

FIGURE 2. S-SO42-/S-SO2 as a function of SO2 concentration. SGHA and SGHB are the silica gel tubes in the second sampling train. 2. As shown, the maximum was 0.13 and it occurred at a very low SO2 concentration. At high SO2 concentrations, which are the focus of this study, they were below 2%. The low ratios demonstrate that the honeycomb denuder system can be used for measuring SO2 concentration accurately and the interference of particulate sulfate is negligible. Aerosol Loss of HDS. The aerosol loss in the HDS causes an underestimate of sulfate concentration. Two mechanisms causing the aerosol loss in the HDS are (1) the aerosol with size larger than 10 µm can be removed by the impactor of the HDS, and (2) the aerosol can diffuse to the wall of the denuder. The aerosol concentration with size-resolved information from the cascade impactor (CI) can be employed to correct the first mechanism. The sulfate diffusional loss, [SO24 ]DF, can be calculated by eq 3. [SO24 ]DF )

∑ DF × [SO i

24 ]CIi

(3)

i

where DFi is aerosol deposition fraction of aerosol size range i and [SO24 ]CIi is sulfate concentration of aerosol size range i measured by the CI. The aerosol deposition (DF) in the honeycomb denuder due to diffusion can be calculated according to eq 4ba and b (28). DFi ) 5.5µi2⁄3 - 3.77µi for µ < 0.009

(4a)

DFi ) 1 - 0.819 exp(-11.5µi) - 0.0975 exp(-70.1µi) for µ g 0.009 (4b) where µi ) DiLN/Q; µi is a dimensionless deposition parameter for particle size range i; Di is diffusion coefficient of the particle size range i (cm2/min); L is the length of the tube () 9.6 cm); N is the number of tubes () 160); and Q is the volume flow rate through the tube () 500 mLpm).

Table 2 shows the mean ratio and its range of the sulfate diffusional loss, the sulfate loss due to the HDS’ inner impactor ([SO24 ]H2SO4 > 10 µm, derived from the CI measurement) and the total sulfate loss ([SO42-]HDloss ) [SO42-]H2SO4>10 µm + [SO42-]DF) to the total sulfate concentration from the CI ([SO42-]CI) at each plant. The mean sulfate loss due to the aerosol diffusional loss at each plant ranged from 3.2% to 6.3%. For all plants, the mean loss was 4.8% demonstrating that the effect was minor. On the other hand, the sulfate loss due to the impactor, shown in Table 2, was highly significant. The mean loss ranged from 3% to 43% demonstrating that the loss varied remarkably, which depends on the size distribution of the aerosol. If the HDS impactor sample were analyzed in this study, the information could be used to directly correct the majority of the loss. The CI information is then no longer needed. Sulfur Dioxide Adsorption. Figure 3 shows the artifact sulfate concentration as a function of SO2 concentration. As feed SO2 concentration increased, the artifact sulfate concentration increased. This is because more SO2 can be adsorbed by the silica gel and the glass fiber filter, and both were not saturated yet. The relation between the feed SO2 concentration (1.6-5.6 ppm) and the artifact sulfate concentration can be described by eq 5. [SO24 ]t ) 20.04 × [SO2]o + 53.32 for 1.6 ppm < [SO2]o < 5.6 ppm (5) 3 where [SO24 ]t is the artifact sulfate concentration (µg/m ); and [SO2]o is the feed SO2 concentration (ppm). In this study, this equation was applied for SO2 concentrations between 1.6 and 5.6 ppm while the deactivation model was applied for SO2 concentrations from 0.2 to 1.6 ppm. The intercept in eq 5 might be from the residual sulfate to be discussed in the next section. Since the ratio of the mean amount of residual sulfate of the second silica gel section to that of the first silica gel section was 2.1, shown in Table 3, outliers were defined as those with ratios exceeding 2.1. Theoretically, the first section should collect more SO2 gas than the second section. However, the high residual sulfate artificially yields a higher sulfate concentration in the second section. Since the mean ratio of the sulfate concentration in the second section to that of the first section is 2.1, outliers are defined as exceeding this ratio. To determine the effect of humidity, experiments were conducted for the relation between adsorbed water amount and sampling time. RHs of 35% and 85% for various sampling times and flowrates were investigated. It was found that the silica gel tube reached saturation quickly. Under these sampling flowrates and sampling times, the adsorbed water amount was around 0.2 g which is one-third of the silica gel weight. When 0.2 Lpm flowrate and RH of 35% were applied, the adsorbed water amount reached saturation in two hours. The results demonstrate that the sampling tube is generally operated under the saturation condition over a wide range of ambient RH.

TABLE 2. Mean and Its Range of the Sulfate Loss to the Total Sulfate Concentration at Each Plant

A B C D E F-1 F-2 G total

5696

9

[SO42-]DF/[SO42-]CI (%)

[SO42-]H2SO4>10 µm/[SO42-]CI (%)

[SO42-]HDloss/[SO42-]CI (%)

[SO42-]CI (µg/m3)

4.8 (4.0-5.3) 3.2 (2.5-3.9) 4.0 (3.0-5.4) 6.3 (5.9-6.6) 4.8 (4.0-5.4) 4.8 (4.8-4.8) 4.7 (4.6-4.8) 5.7 (5.3-6.3) 4.8 (4.0-5.3)

15.7 (8.3-26.8) 42.8 (31.6-52.6) 26.3 (16.7-33.6) 2.8 (1.3-6.6) 15.0 (4.4-34.9) 13.4 (10.3-16.4) 11.6 (10.7-12.5) 5.8 (1.0-11.3) 15.7 (8.3-26.8)

25.4 (19.0-34.8) 49.1 (39.4-57.7) 34.4 (22.7-44.4) 15.4 (14.4-18.4) 24.7 (14.6-42.9) 23.0 (19.9-26.1) 20.9 (20.2-21.7) 17.3 (13.7-21.9) 25.4 (19.0-34.8)

92.4 (53.2-142.6) 25.7 (21.1-32.3) 31.0 (21.1-43.5) 85.3 (21.3-134.1) 32.0 (5.6-61.3) 16.7 (11.7-21.6) 578.2 (412.7-686.3) 44.3 (10.3-108.7) 92.4 (53.2-142.6)

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 15, 2008

FIGURE 3. Artifact sulfate concentration as a function of SO2 concentration.

TABLE 3. Statistical Results of the Residual Sulfate Concentrations (in µg) of Silica Gel Tubes mean standard deviation max min 1st batch (N ) 10) glass fiber filter 3.1 2.3 silica gel - first section 5.2 5.6 silica gel - second section 10.8 23.9 2nd section/1st section 2.1 4.3

9.0 18.3 78.1 4.3

1.1 0.7 0.9 1.3

2nd batch (N ) 8) glass fiber filter 0.6 0.6 silica gel - 1st section 2.0 0.5 silica gel - 2nd section 1.3 0.6 2nd section/1st section 0.6 1.4

1.9 2.6 2.8 1.1

0.0 1.2 0.9 0.7

Residual Sulfate in Silica Gel Tube. Due to the nonzero intercept in eq 5, 18 silica gel tubes as received were analyzed for residual sulfate concentration. The statistical results of the residual sulfate concentrations in different sections for two batches purchased from the vendor are shown in Table 3. The first batch was used for the experiment conducted in this research while the second batch was purchased in 2004 and used for a prior study (2). For the first batch, the mean residual sulfate concentrations from the glass fiber filter, and

first and second sections of silica gel in the aliquots were 3.09, 5.23, and 10.77 µg, respectively, which are equivalent to 12.9, 21.8, and 44.9 µg/m3 (in the air) at the sampling conditions of 8 h and 0.5 Lpm. Added together, the mean total sulfate concentration due to the residual was 79.5 µg/ m3. For the maximum residual sulfate concentration, the artifact sulfate concentration in the air was 439.3 µg/m3. The total sulfate concentration in the air corresponding to the standard deviations of the sampler was 132 µg/m3. This is higher than the mean residual sulfate concentration, indicating that the residual sulfate concentrations were quite variable. In contrast, the residual sulfate of the second batch was much lower than that of the first batch. Its mean total residual sulfate concentration was only 15.9 µg/m3. In summary, the residual sulfate concentration can contribute greatly to the artifact sulfate concentration in the air, and the impact of variation of residual sulfate in the silica gel tube should not be ignored. Therefore, NIOSH Method 7903 is not suitable for low sulfate concentration sampling, e.g., ambient air sampling. Minimization of Artifact Sulfate. Two methods were applied to minimize the artifact sulfate and their effectiveness was evaluated. The first method was the SO2 adsorption model described in eqs 1 and 5. This method is based on using the known SO2 concentration to calculate the artifact sulfate concentration. In this study, the SO2 concentration was determined by the denuder as discussed earlier. The corrected sulfate concentration based on the model, [SO24 ]model, can then be calculated according to eq 6: 22[SO24 ]model ) [SO4 ]SG - [SO4 ]t

(6)

where [SO24 ]SG is the sulfate concentration from the SG, which includes artifact sulfate. The second method used to minimize artifact sulfate was the HDS (sampling train 2). However, it has the potential to remove true sulfate aerosol. Therefore, the sulfate concentrations from SGHA and SGHB need to be corrected and the correcting equation is shown in eq 7. 22[SO24 ]SGHAC(or SGHBC) ) [SO4 ]SGHA(or SGHB) + [SO4 ]HDloss (7)

Figure 4 shows the sulfate concentrations from the SG, the SGHC (the average of the corrected SGHA (SGHAC) and the corrected SGHB (SGHBC)), the model, and the CI as a function of SO2 concentration. The sulfate concentrations from the SG were always higher than those from other samplers. For most samples, the sulfate concentration from the CI was the lowest and the sulfate concentrations from the model and SGHC were between the SG and the CI. For low SO2 concentration, the sulfate concentrations from the SG and the model were similar due to the expected small artifact sulfate produced. The relative error (E) shown in eq 8 was applied to evaluate the difference and the results are shown in Table 4.

E)

FIGURE 4. Comparison of sulfate concentrations from CI, SG, SGHC, and Model: (a) SO2 concentration lower than 3 ppm, (b) SO2 concentration higher than 3 ppm.

|

|

2[SO24 ]sampler - [SO4 ]CI

[SO24 ]CI

(8)

[SO24 ]sampler is the sulfate concentration measured by SG, SGHAC, and SGHBC. The relative errors of the NIOSH method and the model were relatively large when SO2 concentration was lower than 1.6 ppm; their mean values were 4.46 and 3.30, respectively. When high SO2 concentration, ranging from 1.6 to 5.6 ppm, was observed, both the model and the honeycomb denuder reduced the sulfate concentration, making them closer to the CI results. Relative errors of the honeycomb denuder and the model were low at the high concentration range. However, it should be addressed that the artifact sulfate VOL. 42, NO. 15, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5697

TABLE 4. Relative Error of Four Samplers SG

SGHAC

SGHBC

model

mean SDa min max

[SO2] < 1.6 ppm (N ) 21) 4.46 1.73 1.86 2.12 1.59 1.48 0.47 0.06 0.14 8.56 5.76 5.49

3.30 2.23 0.07 7.87

mean SD min max

1.6 ppm